Mössbauer-Effect Studies on Metal Binding in Purine Compounds1a

John I. Peterson , Seth R. Goldstein , Raphael V. Fitzgerald , and Delwin K. Buckhold. Analytical Chemistry 1980 52 (6), 864-869. Abstract | PDF | PDF...
0 downloads 0 Views 1019KB Size
4346

Mossbauer-Eff ect Studies on Metal Binding in Purine Compounds’” I. N. Rabinowitz, Frank F. Davis, and Rolfe H. Herberlb Contribution .from the Department of Physiology and Biochemistry and School

of Chemistry, Rutgers, The State University, New Brunswick, New Jersey. Receiued January 17,1966 Abstract: The nature of the binding site of transition metals in nucleotides and nucleic acids is still largely unsettled.

Mossbauer effectspectra have been obtained on FeOI) and Fe(II1) chelates of nucleotides, nucleic acids, and EDTAtype chelating agents. Spectra have also been obtained for Sn(1V) chelates of nucleotides. The spectra of the Fe(1II)-EDTA-type chelates are analyzed with respect to a previously offered suggestion of an experimental Mossbauer effect test for covalency. The nucleotide and nucleic acid spectra are similarly discussed in terms of degree of covalency of binding and the geometry of the binding site. The Mossbauer effect data support the view that Fe(I1) and Fe(II1) can bind to the purine nitrogens at neutral and basic pH, but not at acidic pH. The Mossbauer effect results complement quite satisfactorily epr spectra obtained from the same compounds. Correlations with infrared and nmr results are also discussed. mphasis in the study of the structure and reactivity of metal complexes of ATP2 has recently shifted to studies of reaction kinetics and detailed analyses of the electronic charge configuration of the complexes. have surveyed some of the concepts Fukui, et involved, and Fukui, et ~ l .have , ~ summarized the possible models for metal chelation to ATP, Metal binding to naturally occurring polymers of nucleotides, such as RNA, is known to affect the conformation and physical properties of the RNA m ~ l e c u l ealthough ,~ it is not yet clear whether the naturally occurring heavy metals in RNA have any effect upon the biological function of the molecule.6-8 Cohn and Hughesg have studied the proton and phosphorus nuclear magnetic resonance (nmr) spectra in metal complexes of ADP and ATP. They found that some metals bind only to the p and y phosphates of ATP and some t o the CY, p, and y phosphates, and that zinc may bind to the nitrogen(s) of the purine ring as well as t o the p and y phosphates. Alteration of pH was shown to affect the position of the resonance peaks. Fe(I1) also binds to the purine rings as well as to the phosphate chainlo Eisinger, et al.,” deter-

E

(1) (a) This work was supported by Grant GM-0999, U. s. Public Health Service, and in part by the U. S. Atomic Energy Commission under Contract AT(30-1)-2472. We thank the Department of Chemistry, Columbia University, for the use of their epr spectrometer, and Mr. Shawn Shih for performing the actual runs. Mr. Hans Stockler of the Department of Chemistry, Rutgers University, was a valuable consultant to I. N. R. on Mossbauer-effect technology. This paper constitutes document NYO-2472-39, U.S.A.E.C. (b) National Science Foundation Senior Postdoctoral Fellow, 1956-1966. (2) Abbreviations used in this paper are AMP, ADP, ATP for adenosine mono-, di-, and triphosphate, respectively. RNA = ribonucleic acid. Other abbreviations used are given in the Experimental Section in parentheses, after the full term appears. (3) K. Fukui, K. Morokuma, and C. Nagata, Bull. Chem. Soc. Japan, 33, 1214 (1960). (4) K. Fukui, K. Morokuma, and C. Nagata, ibid., 36, 1450 (1963). (5) K . Fuwa, W. E. C . Wacker, R . Druyan, A. F. Bartholomay, and B. L. Vallee, Proc. Natl. Acad. Sci. U.S., 46, 1298 (1960). (6) W. E. C. Wacker and B. L. Vallee, J . B i d . Chem., 234, 3257 (1959). (7) R. W. Holley and V. A. Lazar, ibid., 236, 1446 (1961). (8) W. E. C. Wacker, M. P. Gordon, and J . W. Huff, Biochemistry, 2, 716 (1962). (9) M . Cohn and T.R. Hughes, Jr., J. Biol. Chem., 237, 176 (1962). (LO) Personal communication from Professor M. Cohn; methods used are given in ref 9. (11) (a) J. Eisinger, R. G. Shulman, and W. E. Blumberg, Nature, 192, 963 (1961); (b) J. Eisinger, R. G. Shulman, and B. M. Szymanski, J .

mined nuclear relaxation times of protons in metalDNA complexes and concluded that whereas Mn(II), Cu(lI), and Cr(I1) are bound to exterior phosphate groups, Fe(II1) is bound to interior purine coordination sites. Singer12 studied the firmness of binding of various metals to tobacco mosaic virus R N A by repetitive precipitations in alcohol and concluded that Fe(III), among others, binds to the bases, whereas Ca(II), among others, binds only to the phosphates. Maling, et al.,13 found that Mn(I1) and Fe(II1) gave similar electron paramagnetic resonance (epr) signals in complexes with ATP and RNA suggested that Mn(I1) may be bound to the base or ribose moiety as well as the phosphates in these compounds. Patten and Gordy,14 however, found that the epr signals from Mn(I1)-RNA coniplexes were similar to those obtained with Mn(1I) salts in solution. The Mossbauer parameters of greatest interest to the chemist, the isomer shift (IS) and quadrupole splitting (QS), are sensitive to changes in nearest-neighbor interactions with the Mossbauer atom. Since at least one of the binding sites for Fe(II1) in nucleotides is thought to be the polyphosphate chain, the Mossbauer spectra of Fe2(P01)3,Fe4(P20,)3,and Fe(II1)-adenine nucleotides were studied in detail. In addition, since the possible models for metal chelation by nucleotides include structures with four or five oxygen ligands together with two or one bonding interactions with nitrogen atoms, respectively,4 a series of model iron chelates having these structures was studied uia their epr, infrared, and Mossbauer spectra. The model compounds chosen were of the “EDTA family”; Le., nitrilotriacetate (NTA), ethylenediaminetetraacetic acid (EDTA), diethylenetriaminepentaacetic acid (DTPA), and cyclohexanedianiinetetraacetic acid (CDTA). In the order given, the Fe(II1) us. Fe(I1) stability constant increases and the oxidation-reduction potential decreases in the pH range 5-9, which indicates that the Chem. Phys., 36, 1721 (1962). (c) After submittal OF this paper further nmr studies of metal binding to nuclcotides and nucleic acids were reported in a series of papers by R. G. Shulman, e t al., J . Chem. Phys., 43, 3116, 3750 (1965). (12) B. Singer, Biochim. Biophys. Acra, 80, 137 (1964). (13) J. E. Maline. -, L. T. Taskovich, and M. S . Blois, Jr., BioPhys. J . , 3, 7911963). (14) R. A. Patten and W. Gordy, Nature, 201, 361 (1964).

Journal of the American Chemical Society / 88:19 / October 5, 1966

4347

Fe(II1) chelate in these compounds is stabilized relative to the Fe(I1) chelate.16 This provides an independent measure of “covalency” against which another proposed experimental test for “covalency” based on Mi ssbauer analysis l6 was investigated, as discussed be ow. Results and Discussion As shown in Table I, there is no quadrupole splitting in the Fe(II1) nucleotides at pH 1 to 3 except for sample 5 which will be discussed below, or in Fe4(P207)3;in contrast, the QS in ferric phosphate is easily resolved from the resonance spectra. The crystal symmetry of Fe2(P04)3is D 3 4or D36,17for which the screw axis does not allow perfect octahedral symmetry around the Fe(II1) atom. This noncubic nearest-neighbor environment gives rise to a finite electric field gradient which is experimentally observed as a splitting of the first excited nuclear energy level in Fe57. The isomer shift and quadrupole splitting values for Fe2(P04)3,as shown in Figure l a and Table I, are in good agreement with those reported by Fluck, et aZ.’* The crystal symmetry of Fe4(PZ0&has not been reported in the literature, but early crystallographic work on tetravalent metal salts (including Sn(1V)) of pyrophosphoric acidIg and a more recent report on Na4Pz0,20lead to the expectation that the iron atom in Fe4(P20& has six nearest-neighbor oxygen ligands in octahedral symmetry. This symmetry should result in a zero electric field gradient tensor and hence to the absence of quadrupole splitting. The Mossbauer spectrum of Fe4(P20&, shown in Figure 1b, has a large resonance effect, and there is no resolved quadrupole splitting although the line width is about 2.5 times greater than the expected upper limit for an Fe57absorption line.21For the nucleotides at low pH, then (except perhaps for sample 5), the Mossbauer spectral data support the view that the Fe(II1) atom is surrounded by an octahedral arrangement of oxygens, although it is not immediately possible to say which of these oxygens are bound to phosphorus atoms and which to hydrogen atoms in water molecules. The isomer shift values for all of the nucleotides at all p H values, and for all of the chelate model compounds, are in the range expected for ionic Fez+ or Fe3+ compounds. For discussions of the Mossbauer systematics of iron compounds, see, for example, Wertheim,z2 or Gol’danskii. 2 3 Among the chelate compounds studied, however, there is available chemical evidence of covalent bonding interactions, and the Mossbauer spectral parameters of these compounds were examined to see if any systematic qualitative differences could be observed which could lead to quantitative descriptions of the bonding involved. Until this latter is achieved use is made in this paper of the more noncommittal Fe(I1) or Fe(II1) notation. (15) J. Bond and T. I. Jones, Trans. Faraday SOC.,55, 1311 (1959). (16) L. M. Epstein, J . Chem. Phys., 40, 435 (1964). (17) E. C. Shafer, M. W. Shafer, and R. Roy, Z . Krist., 108, 263 I‘ 1956). (18j E. Fluck, W. Kerler, and W. Neuwirth, Angew. Chem. Intern. Ed. Engl., 2, 277 (1963). (19) G. R. Levi and G. Peyronnel, Z.Krist., 92, 190 (1935). (20) D. M. MacArthur and C . A. Beevers, Acta Cryst., 10, 428 ~

(19571 \___.,.

(21) U. Gonser and R. W. Grant, Biophys. J . , 5, 823 (1965). (22) G. K. Wertheim, “Mossbauer Effect, Principles and Applications,” Academic Press Inc., New York, N. Y., 1964. (23) V. I. Gol’danskii, “The Mossbauer Effect and Its Applications in Chemistry,” Consultants Bureau, New York, N. Y., 1964.

P”

~

-20

-IO

0

IO

20

30

Velocity, mmisec. Figure 1. Mossbauer spectrum of (a) Fen(PO&, 300°K; (b) Fe477°K.

(p207)8,

For the EDTA series of Fe(II1) chelates, which have similar binding site symmetry and nearest-neighbor ligand atoms, systematic IS values were found. I t has been suggestedl6 that as the 3s character (“covalency”) of an iron atom increases within a series of closely related compounds, a shift toward more negative values for the IS parameter should be observed. As shown in Table I, samples 23 to 28, this seems generally to be the case for the series Fe(II1)-NTA, -EDTA, -DTPA, -CDTA, where independent evidence15 indicates that the covalent character of the iron atom increases in the order given. Some of the dangers of a too facile interpretation of Mossbauer data, as well as some of the interesting potentialities for this technique, may be seen in a closer analysis of the EDTA family of chelates. NTA, for example, is known to have a slight chelating tendency toward sodium ionsz4and this may be reflected in the different isomer shifts of Fe(II1)-NTA when potassium or sodium are the counterions. The different isomer shifts could be due to the competitive effect of sodium bonding forcing the iron to adopt more than one binding site so that the Mossbauer spectrum would consist of two superposed resonances effectively changing the observed IS and QS values. It seems less likely that the IS value, arising from the electronic charge density at the iron nucleus, would be differentially affected by an electrostatic field of sodium as opposed t o potassium ions. The former effect should be concentration dependent and it was found that within a wide range of sodium concentration there was no appreciable difference in the Mossbauer parameters of Fe(II1)-ATP at low pH (samples 6 and 7, Table I). Sodium was therefore used as a convenient counterion in all of the other complexes investigated. When the ATP :Fe(II1) molar concentration ratio was increased from 1 :1 to 10: 1 a t low pH, a surprising bumpy spectrum was obtained, as seen in Figure 2 (sample 8). A similar spectrum showing unresolved fine structure, with more than one major resonance, was observed by Epstein for Fe(II1)-EDTA, and it was (24) G. Schwarzenbach and W. Biedermann, Xelu. Chim. Acta, 31, 331 (1948).

Rabinowitz, Davis, Herber 1 Metal Binding in Purine Compounds

4348

Table I

3 Fe(II1)-AMP 4 Fe(II1)-ADP, Chelex-treated 5 Fe(II1)-ADP 6 Fe(II1)-ATP

7 Fe(lI1)-ATP

8 Fe(II1)-ATP 9 Fe(II1)-ATP

10 Fe(II1)-ITP 11 Fe(II1)-ATP in 70% glucose 12 Fe(II1)-ATP 13 Fe(II1)-ATP in 70% glucose 14 Fe(II1)-ADP 15 Fe(II1)-AMP ~

"K

Line width, mm/sec 1.16

IS,"

QS, mm/sec

Concn, M

PH

Fe = 1 . 8 X NaaPz07= 1 . 8 X ppt obtained Ppt with 1 M NaH2P04and FeCI3.6Hz0 Fe = 3.0 X AMP = 3 . 6 x 10-3 Fe = 2 . 4 X ADP = 3.7 x 10-3 Fe = 8.3 X ADP = 1.2 X 1W2 Fe = 2.3 X ATP = 2.8 x 10-3 ionic strength (NaCI) = 1.0 Fe = 6 . 3 X ATP = 7.2 x 10-3 ionic strength, (NaC1) = 3.5 X Fe = 1.8 X ATP = 1.8 X Fe = 1.3 X ATP = 8.3 x 10-3 Fe = 1.3 X ITP = 8.3 x 10-3 Fe = 9.4 X ATP = 1.1 X Fe = 1.8 X IO+ ATP = 1.8 X Fe = 1.2 X ATP = 1.2 X Fe = 1.3 X ADP = 1.1 X Fe = 6.3 X l W 4 AMP = 8.0 x 10-3

3.0

77

2.0

300

1.9

77

0.92

0.53

2.0

77

1.14

0.81

1.8

77

3.2

77

1.36

0.71

1.6

77

1.12

0.71

1.6

77

1.13

0.64

7.1

77

0.70

0.61

7.1

77

0.75

0.51

6.0

77

0.71

0.69

9.8

77

0.68

0.48

9.7

77

0.74

0.69

7.1

77

0.88

0.63

7.1

77

0.79

0.74

Sample

2 FedPO4h

Temp,

mmlsec 0.70 0.69

0.70

0.66

0.48

~~~

a As stated in the Experimental Section, the isomer shifts for the iron samples are with respect to a standard absorber of Na2Fe(CN),N0. 2Hz0. For (Pd) sources (used in this work), this amounts t o adding 0.442 mm/sec to each IS value. * No independent determinations of molecular weight were made for these molecules. An estimated molecular weight for poly A (Calbiochem) would be 200,000 and for

suggestedzs that there were two or more discrete binding environments for the Fe(II1) in EDTA. Other examples of such complex spectra, exhibiting hyperfine (nuclear Zeeman) interactions between the magnetic

n

I

-30

-20

I

-IO

I .

1

0

10

I

1

20

30

4.0

50

Velocity, mmisec. Figure 2. Mossbauer spectrum of Fe(II1)-ATP, pH 1.6, 77'K; the molar ratio of nucleotide to Fe is 1O:l.

moment of the iron nucleus and the surrounding magnetic field, have been reported, notably in ferrichrome and ferrichrome A by Wickman, et al. (quoted (25) L. M. Epstein, J . Chem. Phys., 36, 2731 (1962).

Journal of the American Chemical Society

88:19

by Gonser and Grantz6). A theoretical treatment of the effect on Mossbauer resonances of fluctuating electric and magnetic fields produced by the relaxation of paramagnetic ions or by the fluctuation of the environment surrounding the nucleus has been reported by B l ~ m e . ~The ' spectrum of ferrichrome A at 77"KZ6 closely resembles the spectrum shown in Figure 2, and in the case of ferrichrome A the temperature dependence of the hyperfine splitting indicated that the fluctuations in the internal magnetic field were due to a long electronic spin-lattice relaxation time affecting the Mossbauer resonance.26 In the case of sample 8 of Fe(IIIFATP, the concentration dependence of the hyperfine structure indicates that the fluctuations are due to electronic spin-spin relaxation, although the temperature dependence has not been checked. The ATP :Fe(II1) concentration ratio of the complex at pH 7.1 (sample 9) is not exactly the same as that at pH 1.6 (sample 8), but there is no trace of hyperfine splitting. This supports the argument given below that the binding sites are different at these two pH values. A further complication in the analysis of the Fe(II1)-EDTA Mossbauer spectrum, and, by implication, the other chelates of this family, is X-ray diffraction evidence in (26) U. Gonser and R. W. Grant, "Mossbauer Effect Methodology," Vol. I, Plenum Press, New York, N. Y.,1965,pp 36-37, (27) M. Blume, Phys. Reo. Letters, 14,96 (1965).

October 5, 1966

4349

Sample 16 Fe(II1)-poly A 17 Fe(II1)-Me-poly A 18 Fe(II1)-S-RNA 19 Fe(I1)-ATP 20 Fe(I1)-ATP 21 Fe(I1)-ADP 22 Fe(I1)-AMP 23 Fe(1II)-NTA potassium salt of NTA 24 Fe(II1)-NTA sodium salt of NTA 25 Fe(II1)-NTA 26 Fe(II1)-EDTA 27 Fe(II1)-CDTA 28 Fe(II1)-DTPA 29 FerTr(A-H) 1 2 30 Fe”’B(0H) (ClO& 31 Sn(1V)-AMP Mg2Sn source 32 Sn(1V)-ADP Mg2Sn source 33 Sn(1V)-ATP Mg2Sn source 34 Sn(1V)-ATP Sn02 source/

Concn, M

PH

1 .O mg of Fe 102.0 mg of poly Ab 0 . 7 mg of Fe 75.0 mg of Me-poly Ab 1.0mgofFe 101 .O mg of S-RNAb Fe = 1.7 X 10-3 ATP = 1.7 X loV2 Fe = 1.4 X ATP = 1.4 X Fe = 2.0 X 10-8 ADP = 2.0 X Fe = 1.8 X AMP = 1.9 X Fe = 6.1 X NTA = 6.7 X

7.0

Temp,

Line width, mm/sec

IS,a

QS,

mm/sec

mm/sec

77

0.72

0.74

7.0

77

0.77

0.79

7.2

77

0.63

0.68

1.9

77

1.58

3.10

7.2

77

1.8

77

(i) 0. 56c (ii) 1.77 1.61

0.72 2.42 2.90

3.0

77

1.53

3.02

6.0

77

0.77

1.39

Fe = 1.0 X NTA = 1.2 X

5.3

77

0.55

1.68

Fe = 1.0 X NTA = 1.2 X

9.1

77

0.66

0.70

7.3

300 77

7.0

77

0.39

0.56

1.00

0.79 0.80 -1.79

1,62 0.85

1.5

77 77 300

1.5

300

1.10

-1.88

1.6

300

1.16

-1.89

1.6

77

1.44

-0.04

. . .d Fe = 7.8 X CDTA = 4 . 4 X Fe = 7.1 X 10-8 DTPA = 4.0 X 10-2

. . .d . . .6 SN = 1.5 X AMP = 1.7 X 10-2 Sn = 1.4 X ADP = 1.6 X Sn = 1.1 X 10-2 ATP = 1.2 X 10-2 Sn = 1.1 X 10-2 ATP = 1.2 X 10-2

OK

~~

1.82

0.62 0.62

~~

S-RNA, 25,000. Interpretation of the two resonance effects given in text. d Cf ref 29. 8 C ’ ref 28. f A SnOz source was substituted for the MaSn source used for samples 31 to 33 after the Mg2Snsource disintegrated in midrun for sample 34. The velocity conversion factor for Mg2Sn source to a Sn02source is +1.76 mmisec.

favor of seven rather than six coordination of the iron atom. 2a Because the geometry of the binding site for Fe(II1) in nucleotides is not definitely known, and because electric field gradients in ionic Fe3+ compounds are very sensitive to small changes in bond lengths and angles as well as t o postulated ligand charges, 2 2 crystalfield models of the binding sites, yielding information about the strength of the ligand field and values for crystal-field splittings, are not advanced here. Once the molecular parameters are accurately known, quantitative statements about the extent of the deviation from a purely ionic Fe3+ nucleotide structure may be made from the Mossbauer data. It is possible, however, t o detect changes in the electric field gradient in a given compound when one ligand is substituted for another. The difference, for example, in the QS between Fe(II1)NTA at pH 5.3 and pH 9.1 (samples 24 and 25, Figure 3), may be due to one aquo ligand being replaced by a hydroxo ligand. 2 4 , 2 9 The structural differences between the pH 5.3 and pH 9.1 Fe(II1)-NTA chelates can also be observed in the infrared spectra of the KBr pellets of the powdered solids, 3O although a quantitative treat(28) J. L. Hoard, M. Lind, and J. V. Silverton, J . A m . Chem. Soc., 83, 2770 (1961). (29) R. I. Gustafson and A. E. Martell, J . Phys. Chem., 67, 576 (1963).

ment of the infrared spectra was not attempted in view of the difficulties involved, as discussed by Nakamoto. 3 1 Epr spectra of the two chelates showed only marginal differences, which is perhaps to be expected with powder samples. The absence of resolved quadrupole splitting in Fe(II1)-CDTA (sample 27) is a further caution not to equate “chemical” with electric field symmetry. 3 2 A more complex example of this is that of the Mossbauer spectra of two Fe(II1) chelates, prepared by Curry and B u ~ c h(samples ~~ 29 and 30, Figure 4). As with the EDTA family of chelates, these two chelates are presumed to be six-coordinate, but instead of having one or more electronegative oxygen atom ligands replaced by electropositive nitrogen atoms, the situation is reversed. Both chelates are low-spin Fe(II1) chelates and the first sexadentate chelate (sample 29), all the ligands of which are nitrogen atoms, is a highly strained structure with low symmetry33 which would therefore be expected to show quadrupole splitting, as observed. The second chelate (sample 30) (30) I. N. Rabinowitz, Dissertation Abstr., 24, 171 (1964). (31) K. Nakamoto, “Infra-Red Spectra of Inorganic and Coordination Compounds,” John Wiley and Sons, Inc., New York, N. Y., 1963. (32) R. H. Herber and H. A. Stockler, Trans. N . Y . Acad. Sci., 26, 929 (1964). (33) J. D. Curry and D. H. Busch, J . Am. Chem. SOC.,86, 592 (1964).

Rabinowitz, Davis, Herber

I Metal Binding in Purine Compounds

4350

U I

-3.0

I

-1.0

,

-1.0

I

I

,

I

,

,

,

0

10 .

2.0

3.0

4.0

0.0

6.0

Velocity, mmisec. Figure 5. Mossbauer spectrum of (a) Fe(I1)-ATP, Fe(I1)-ATP, pH 7.2 (both at 77°K).

Velocity, mmisec. Figure 3. Mossbauer spectrum of (a) Fe(II1)-NTA, pH 9.1, upper velocity scale and counting range 101,000 to 105,000; (b) Fe(II1)NTA, pH 5.3 (both at 77’K). Note: There are some Mossbauer curves with more than one abscissa and more than one set of numbers giving “counts per channel.” This is because the total counting rate was not always the same for all samples and because the velocity scale (the abscissa) therefore would necessarily be different for compounds which were measured when the apparatus had differing velocity constants. It is nearly impossible to put all of the curves on one standard abscissa.

..

I

I3’t

113

-

130

3.0

-2.0

-10

0

1.0

2.0

3.0

Velocity, mmisec. Figure 4. Mossbauer spectrum of (a) Fe11xB(OH)(C104)2,upper velocity scale and counting range 133,000 to 137,000; (b) FeIII(A-H) I1 (both at 77°K).

differs from the first in the substitution of a hydroxo ligand for an imino nitrogen.a3 The two chelates differ markedly in the magnitude and asymmetry of their quadrupole splittings. Line asymmetry can be an important parameter in deducing binding site symmetry and structure from the Mossbauer spectra. The task lies in clearly distinguishing from among the possible causes of line asymmetry. There is first the preferential orientation of the polycrystalline sample in the absorber holder;a4 then there is the Gol’danskii effect,S5in which bond strengths stronger in one direction than another give rise to an (34) M. Kalvius, U.Zahn, P. Kienle, and H. Eicher, Z . Naturforsch., 17a, 494 (1962). (35) V. I. Gol’danskii, E. F. Makarov, and V. V. Khrapov, Phys. Reu. Letrers, 3, 344 (1963).

Journal of the American Chemical Society

/

88:19

pH 1.9; (b)

anisotropic Debye-Waller factor, or recoil-free fraction. This effect is probably responsible for the asymmetry of the Q S in samples 29 and 30. The last cause of line asymmetry would be unresolved nuclear Zeeman interactions as mentioned above. Definite identification of the correct cause of line asymmetry will depend on concentration and temperature studies of a polycrystalline sample and on obtaining Mossbauer spectra from single crystals. With the results of the model chelates in mind, we may try to rationalize the increase in the values of the isomer shift for the series AMP, ATP, ADP, at low pH as follows. The iron atom electron orbitals in an octahedral ligand arrangement are hybridized as d 2 sp3 and the two d orbitals used are the eg orbitals in the nomenclature of Orgel, 3 6 the remaining tZga6d orbitals accommodating the electrons originally present on the metal. If Fe(I1I) is bound to a diphosphate moiety in ADP and ATP, the electronegative phosphoryl oxygen atoms can donate electrons t o the tzgorbitals of the iron. This addition of d electrons has a shielding effect on the nucleus which reduces the attractive Coulomb potential between the nucleus and the s electrons. This gives rise to 3s wave function expansion with a concomitant decreased s electron density at the nucleus. This decrease in /9,(0)/ results in a positive change in the isomer shift37 in ADP and ATP relative to that of AMP, which cannot bind iron to a diphosphate, except in an intermolecular complex. The increase in isomer shift in the series Fe(II1)AMP, -ATP, -ADP (samples 3, 4, and 6) a t low pH, is repeated for the series Fe(I1)-AMP, -ATP, -ADP (samples 22, 19, and 21), although the data are not quite as clear-cut. The isomer shifts for the Fe(I1) chelates are characteristically larger than those for the Fe(II1) chelates, since the extra d electron in Fe(I1) effectively decreases the s electron density in a manner similar to that discussed above. 37 The quadrupole splitting for the Fe(I1) chelates is also characteristically larger than for the Fe(II1) chelates, as can be seen by comparing Figure 5a with Figure 6b (samples 19 and 9). This observation, which is consistently noted for analogous Fe(I1) and Fe(II1) inorganlc compounds, arises from the fact that the quadrupole splittings for (36) L. E. Orgel, “An Introduction to Transition-Metal Chemistry: Ligand Field Theory,” Methuen and Co., Ltd., London, 1963. (37) L. R. Walker, G. K. Wertheim, and V. Jaccarino, Phys. Reo. Letters, 6, 98 (1961).

/ October 5, 1966

4351

I 116

Velocity, mmisec.

1

-30

Figure 6. Mossbauer spectrum of (a) Fe(II1)-ATP, pH 1.6 (sample 7 of Table I), upper velocity scale and counting range 53,000 to 74,000; (b) Fe(II1)-ATP, pH 7.1 (both at 77°K).

Fe(II1) complexes are due to asymmetric field gradients around the spherically symmetric charge distribution on the Fe(II1) ion, whereas the extra d electron present in complexes of Fe(I1) introduces an asymmetric charge field on the iron atom itself. The energetic state of this extra orbital is determined by the crystal field. When the pH of the Fe(II1) chelates of AMP, ADP, and ATP is brought up to pH 7 (samples 15, 14, and 9), quadrupole splittings are observed. These data strongly suggest that one or more nitrogens of the ring have replaced oxygen ligands; i.e., at low pH the Mossbauer data support a chelate model involving only the phosphate moieties of adenine nucleotides, but at neutrality and higher pH the ring nitrogens have become effective ligands. Nmr evidence seems to point to either the 6-amino and/or 7-imino position on the adenine ring as possible binding ~ites.~~'O An Fe(II1) chelate of inosine triphosphate was prepared in exactly the same manner as sample 9 of Fe(II1)-ATP (sample 10, Figure 7a). The isomer shifts of the two samples were about the same within experimental error, and the QS values were marginally different. The asymmetry of the QS for the two cases appears to be different although the counting statistics and resolution of the Fe(II1)-ITP resonance is not particularly good. In view of the results with the model chelates, a difference in line asymmetry would be expected between the two compounds if the 6-amino position is in fact an effective ligand in the Fe(II1)-ATP chelate. If the pH of the Fe(II1)-ATP chelate is raised to very basic values, the nitrogen of the 6-amino position should become an even more effective electron donor, and as seen in Figure 7b,c (samples 12 and 13), with good resolution, the asymmetries of the quadrupole splittings are opposite in sense to that of the Fe(II1)-ITP (Figure 7a) and more pronounced than that of Fe(II1)-ATP at pH 7.1 (Figure 6b). Glucose (70 %), used as a solvent for samples 11 and 13, eliminated the possibility of ferric hydroxide formation and allowed the molar ratio of iron to nucleotide to be increased so that spectra at high pH could be studied over a wider range of concentrations. At approximately neutral pH, the Mossbauer parameters for Fe(II1)-ATP in both water and 70% glucose were

I -2.0

I

I

-1.0

0

'

I

ID

!

2.0

3.0

Velocity, mmisec. Figure 7. Mossbauer spectrum of (a) Fe(II1)-ITP, pH 7.1, counting range 115,OOO to 122,000; (b) Fe(IIIkATP, pH 9.7, in 7 0 x glucose, counting range 141,000 t o 146,000; (c) Fe(II1)-ATP, pH 9.8, in water, counting range 62,000 to 77,000 (all samples at 77°K).

essentially the same, including line asymmetry, which gave some reason for believing that the nearest-neighbor configuration about the iron atom was the same in both cases, at least at neutral pH. When the pH of an Fe(I1)-ATP chelate is raised to about neutrality (sample 20) the spectrum shown in Figure 5b is obtained. The resonance exhibiting quadrupole splitting and centered at 0.56 mmjsec is interpreted as being due to Fe(III), arising probably from air oxidation of Fe(I1) as the pH is raised. The resonance line at about 3.1 mmjsec is then part of the quadrupole splitting doublet of residual Fe(II), the other part of the doublet being at about 0.56 mmjsec. It would be of interest t o repeat this experiment under an inert atmosphere throughout to determine to what extent, if at all, Fe(II1) is the stable valence state in iron nucleotides, as it is in chelates with the EDTA family. l5 The Mossbauer spectra for Fe(II1)-polyadenylic acid (sample 16, Figure 8b) and Fe(II1)-methylated poly A (sample 17) resemble almost exactly the spectrum for Fe(II1)-AMP (Figure Sa), which supports the thesis that the binding sites in these molecules are similar to that in AMP, i.e., a site involving both the phosphate and the adenine ring. Methylated poly A is methylated a t the N1 position on the adenine ring,38 so that these results provide more evidence that the N1 position is not involved in metal complex formation. The spectrum for Fe(II1)-S-RNA shown in Figure 8c (sample 18) exhibits a slightly different IS, QS, and line asymmetry than either Fe(I1I)-poly A or Fe(II1)-Mepoly A, and this is to be expected as the different bases present in S-RNA allow different populations of binding sites and the resultant of the separate Mossbauer resonances should yield an effect having a different IS, Q S , and line asymmetry. Sample 5 of Fe(II1)-ADP at low pH, unlike Fe(II1)AMP or Fe(II1)-ATP at low pH, exhibits a clearly resolved quadrupole splitting (Figure 9a). When the ADP was first passed through Chelex resin to remove (38) D.B. Ludlum, R. C. Warner, and A. J. Wahba, Science, 145, 397 (1964).

Rabinowitz, Davis, Herber

Metal Binding in Purine Compounds

4352

Velocity, mm/sec. Figure 10. Mossbauer spectrum of (a) Sn(1V)-ATP, pH 1.6, MgSn source, 300”K, counting range 304,000 to 310,000; (b) same sample as (a) but with an SnOn source and at 77 “K. Upper velocity scale and counting range 76,000 to 81,000. -30

-20

-10

0

IO

20

30

Velocity, mmisec. Figure 8. Mossbauer spectrum of (a) Fe(II1)-AMP, pH 7.1, counting range 120,000 to 141,000; (b) Fe(II1)-poly A, pH 7.0, counting range 120,000 to 141,000; (c) Fe(II1)-S-RNA, p H 7.2, upper velocity scale and counting range 112,000 to 120,000 (all samples at 77°K).

Velocity, mmisec. Figure 9. Mossbauer spectrum of (a) Fe(II1)-ADP, pH 1.8, not Chelex treated, counting range 35,000 to 37,000; (b) Fe(II1)-ADP, p H 2.0, Chelex treated, counting range 53,000 to 70,000 (both at 77 OK).

approximately two-thirds of the iron “impurity”39 (from 455 to 131 mpg of Fe/mg of nucleotide) present in the commercial sample, and then complexed with added iron, there was no quadrupole splitting (Figure 9b, sample 4) although the isomer shift became more positive. It is possible that the totality of trace element impurities in ADP is responsible for competitive binding with iron, but this seems unlikely in view of the extraordinarily high binding constant found for Fe(II1) to ADP.40 It will be recalled that at low pH, Fe(II1)ATP exhibited concentration-dependent hyperfine interactions, but not quadrupole splittings, and it has also recently been suggested4I that Fe(II1) will form complexes of the form Fe2ATP and Fe(ATP), at pH 2 but that Fe(II1) will form only the 1 : l complex with

ADP. The Mossbauer data would seem to advance one reason for the high binding constant of Fe(1II) to ADP by suggesting that even at low pH the ring nitrogens are involved in chelation. The spectra of the tin nucleotides (samples 3 1 to 34), a representative sample of which is shown in Figure 10 (samples 33 and 34), support the interpretation given for the Fe(II1) nucleotides at low pH. It may also be recalled here that Sn(1V) is known from X-ray diffraction evidence to be in an octahedral binding site in stannic pyrophosphate.1g The existence of a sizable resonance at room temperature for sixcoordinate tin is taken as evidence that the tin is bonded and the to a polymer chain of high molecular present results clearly indicate extensive polymer formation in the tin nucleotides which have been examined. Epr spectra were obtained on many of the same complexes used for Mossbauer analysis. The difficulties of working with powdered samples have been discussed by Maling, et a1.,13 who found that pH and temperature differences changed the spectra of manganese-tripolyphosphate, manganese-ATP, and irontripolyphosphate. Similar spectral changes were observed in this study with iron nucleotides. As shown in Figure 11, for Fe(II1)-ATP, an increase in pH or a lowering of temperature appears to increase the amplitude of the g = 4 peak relative to that of the g = 2 peak. Some of the nucleotides as well as some of the model compounds, e.g., Fe(II1)-CDTA, also showed a clear g = 6 component (Figure 12b), whereas Fe4(P207), showed only a g = 2 component with possibly a ~ trace of g = 4 (Figure 12a). Castner, et u I . , ~ and G r i f T ~ t hhave ~ ~ discussed the meaning of these g values for Fe(II1) complexes, and the conclusions pertinent to the present discussion are that g = 2 will appear when the crystal field is weak, or in an octahedral ligand field when the ligand field is electrically symmetrical. Moreover, g = 4 and g = 6 will appear when the ligand field is made unsymmetrical by the substitution of unequal charges, although the shape of this distorted

(39) For a quantitative survey of the amounts of trace iron found in commercially available nucleotides and a demonstration of a biochemical role for such “impurities,” cf., P. Hochstein, K . Nordenbrand, and L. Ernster, Biochem. Biophys. Res. Commun., 14, 323 (1964). (40) E. H. Strickland and C. R. Goucher, Arch. Biochem. Biophys.,

(42) R. H. Herber and H. A. Stockler, “Panel on the Application of Mossbauer Spectroscopy to Problems of Chemistry and Solid State Phvsics.” IAEA. Vienna. Aoril 1965 (in oress). (43) T. Castner, Jr., G . S. Newell,‘W.-C. Holton, and C. P. Slichter,

108, 72 (1964). (41) C. R. Goucher and J. F. Taylor, J . Biol. Chem., 239, 2251 (1964).

J . Chem. Phys., 32, 668 (1960). (44) J. S. Griffith, Mol. Phys., 8, 213 (1964).

Journal of the American Chemical Society

~

88:19 / October 5, I966

4353

I

9’ 2

,

H

I

2600 A

d

C H 2500

I

L

b-

I

(C) L

1

H 2500

4

I

Figure 12. Epr spectrum of (a) Fe4(P201)3; (b) Fe(1II)-CDTA, pH 7.3 (both at 300°K).

(d) g:2 5

C H 20

Figure 11. Epr spectrum of (a) Fe(II1)-ATP, pH 2.7, 300°K; (b) same sample as in (a), but at 77°K; (c) Fe(II1)-ATP, pH 7.1, 300°K; (d) Fe(II1)-ATP, pH 1.6; molar ratio of nucleotide to Fe is lO:l, 300°K.

octahedron is not uniquely defined by the appearance of g = 4 and g = 6 resonances.44 It is interesting that a pronounced g = 2.5 peak appears in the Fe(II1)ATP chelate a t low pH when nucleotide: Fe(II1) concentration is 10 : 1 (Figure 1Id). This is the sample that exhibited nuclear hyperfine interactions in its Mossbauer spectrum (Figure 2). Although g = 2.5 peaks have been noted in nucleic acids, 4s no explanation has yet been offered which accounts for this observation.

Experimental Section The purest available grades of adenine nucleotides were obtained from Schwarz Bio-Research, Inc., or from the Sigma Chemical Co. Inosine 5’-triphosphate (ITP) was obtained from Mann Research Laboratories, Inc., as the “chromatographically pure” sodium salt. Polyadenylic acid (poly A) was obtained from Calbiochem as the “B” grade potassium salt and used without further purification. Methylated polyadenylic acid (Me-poly A) was prepared from poly A according to the method of Ludlum, et uI.,as for 13.5 % methylation. Soluble ribonucleic acid (S-RNA) was prepared according to the method of Penniston and Doty46 from fresh pressed yeast (Saccharomyces cereuisiue). Nitrilotriacetate was obtained from Distillation Products Industries. Cyclohexanediaminetetraacetic acid and diethylenetriaminepentaacetic acid were gifts of the Dow Chemical Co. The two compounds, Fe”’(A-H)Iz and FeX1’B(OH)(ClO&,whose structures are described in the Results and Discussion section, were gifts from Dr. D. H. (45) W. M. Walsh, Jr., L. W. Rupp, Jr., and H. J. Wyluda, “Paramagnetic Resonance Proceedings of the First International Conference Held in Jerusalem, July, 1962,” W. Low, Ed., Academic Press Inc., New York, N. Y., 1963. (46) J. T. Penniston and P. Doty, Biopolymers, 1, 145 (1963).

Busch of the Ohio State University. Chelex 100, 100-200 mesh, was obtained from Bio-Rad Laboratories. Chelex treatment for nucleotides was carried out at 10’ using a column or batch process. Cot-free KOH was prepared according to the method of Schwarzenbach and Biedermann24 and stored with ordinary precautions, as was NaOH made up COz free by the usual procedure from saturated NaOH and boiled distilled water. Enriched Fe5’ in the form of Fe2O3(85 % Fe57)was obtained from Oak Ridge National Laboratory and converted to FeCl, with concentrated HCl. Enriched Fe57Clz was prepared from a slightly acid solution of FeS7Cl3by bubbling hydrogen gas through the solution for 17 hr over small pieces of platinum wire.47 Analyses for inorganic phosphate were carried out according to the methods of Fiske and Subbarow and Lowry and Lopez.48 The amounts of inorganic phosphate found in the nucleotide chelates were negligible. Analyses for iron content were carried out using the “bathophenanthroline” method,49 slightly modified for use with nucleotides. 89 All complexes studied were prepared as soluble aqueous systems unless otherwise noted. The concentrations of the reagents, in solution, are given in Table I, and the pH given is likewise for the solution. The physical state of the compounds for all of the spectral analyses reported here was a lyophilized powder. Infrared spectra were obtained with a Perkin-Elmer Model 21 recording spectrophotometer with sodium chloride optics. The epr spectra were obtained using a Varian V-4500 X-band spectrometer with 100-Kc modulation and using a Varian Gin. magnet. The parabolic motion Mossbauer spectrometer used in this study has been described in detail e l s e ~ h e r e . 5 ~The ~ ~ ~samples were generally uniform powders less than 1 mm thick pressed between aluminum foil supports. The velocity constant was determined periodically (before or after a sequence of sample determinations) by determining the multi-channel analyzer addresses of the peaks in the magnetic hyperfine spectrum of metallic iron. These data can be used to obtain the Doppler shift constant (mm/sec/ channel) for the spectrometer by using the precision data of Preston, e t ~ 1 . 6 1 The zero of motion was determined independently using a S n 1 W Zsource us. SnOz absorber, both at room temperature. For work at 77”K, the absorber holder was attached to a cooling prong of a dewar flask. The parameters for the iron complexes (47) C. A. Fredenhagen, Anorg. Chem., 29, 405 (1902). (48) L. F. Leloir and C. E. Cardini, Methods Enzymol., 3, 840 (1957). (49) H. Diehl and G. F. Smith, “The Iron Reagents,” G. Frederick Smith Chemical Co., Columbus, Ohio, 1960. (50) R. L. Cohen, P. G. McMullin, and G. K. Wertheim, Reu. Sci. Instr., 34, 671 (1963). (51) F. C. Ruegg, J. J. Spijkerman, and J. R. DeVoe, ibid., 36, 356 (1965). (52) R. S. Preston, S. S. Hanna, and T. Herberle, Phys. Rev., 128, 2207 (1962).

Rabinowitz, Davis,Herber

Metal Binding in Purine Compounds

4354 were obtained using a Co6' source in palladium, and the isomer shift values have been converted to shifts from a standard absorber

pole splittings were measured from center to center of the resolved doublets and the isomer shift of a compound exhibiting a Q S was

of Na2Fe(CN)6N02H20 in order to facilitate comparisons with isomer shift values in the literature.63 To obtain the values for the isomer shifts, the centers of the resonance lines were found using the method of chords.&' Quadru-

determined by averaging the two centers. For well-resolved lines the standard deviation for the IS and Q S is f0.08 mmisec. Curve fitting was not attempted for asymmetric quadrupole splittings with poor resolution.

(53) R.H.Herber in "Mossbauer Effect Methodology," I. J. Groverman, Ed., Plenum Press, New York, N. Y.,1965,p 3.

(54) R. H. Herber and H. A. Stockier, unpublished results; H. Brafman, M. Greenshpan, and R. H. Herber, Nucl. Instr. Methods, in press.

Random-Coil Configurations of cis-1,4-Polybutadiene and cis-1,4-Polyisoprene. Theoretical Interpretation J. E. Mark Contribution,from the Department of Chemistry, The Polytechnic Institute of Brooklyn, Brooklyii, New York 11201. Received May 23, 1966 Abstract: The rotational isomeric state model with neighbor dependence is used to calculate random-coil dimenand 1,4-polyisoprene,+CH2---C(CH3)= sions of the cis forms of 1,Cpolybutadiene, +CH2-CH=CH-CH&,, CH-CH2+,, in the limit of large x. Comparison of calculated and experimental values of the characteristic ratio (r2jo/n12and its temperature coefficient is used to determine intramolecular energies of various conformational sequences of the chain backbone. The form of lowest intramolecular energy closely corresponds to the conformation adopted by these two polymers in the crystalline state.

I

n a series of investigations over the last few years, the rotational isomeric state model has proved remarkably successful in the interpretation of configurational characteristics of chain molecules. An area of particular importance is the analysis of the chain extension, and frequently also its temperature coefficient, for polymer chains in the limit of high molecular weight. Such studies have been carried out for polyisobutylene, polyethylene,?, 3 poly(dimethylsiloxane), p~lyoxymethylene,~polyoxyethylene,6 polypeptide^,^ and vinyl polymers of both tactic and atactic structure.8 The fact that this model has also been shown to give a very satisfactory account of the dipole moments of a,w-dibromo-n-alkanesg and oxyethylene molecules'O in the region of short chain length strengthens confidence in its application. The availability of experimental values of the chain extension and its temperature coefficient for cis-l,4polybutadiene and cis- 1,4-polyisoprene now permits similar analyses of these two polymers, which are of considerable importance because of their relatively simple chemical structure as well as their extensive commercial utilization. The correlation of experimental results with results calculated using a rotational (1) 0.B. Ptitsyn and Yu. A. Sharanov, Zh. Tekh. Khim., 27, 2744, 2762 (1957); C.A.J. Hoeve, J. Chem. Phys., 32, 888 (1960). (2) A. Ciferri, C.A. J. Hoeve, and P. J. Flory, J. Am. Chem. SOC.,83, 1015 (1961); C.A. J. Hoeve, J. Chem. Phys., 35, 1266 (1961); K. Nagai and T. Ishikawa, ibid., 37, 496 (1962). (3) A. Abe, R. L. Jernigan, and P. J. Flory, J . Am. Chem. Soc., 88, 631 (1966). (4) P. J. Flory, V. Crescenzi, and J. E. Mark, ibid., 86, 146 (1964). (5) P.J. Flory and J. E. Mark, Makromol. Chem., 75, 1 1 (1964). (6) J. E.Mark and P. J. Flory, J . Am. Chem. Soc., 87, 1415 (1965). (7) D. A. Brant and P. J. Flory, ibid., 87,2791 (1965). (8) P. J. Flory, J. E. Mark, and A.Abe, ibid., 88, 639 (1966). (9) W. J. Leonard, Jr., R. L. Jernigan, and P. J. Flory, J . Chem. Phys., 43, 2256 (1965). (10) J. E. Mark and P. J. Flory, J. Am. Chem. Soc., 88, 3702 (1966).

Journal of the American Chemical Society

I 88:19

isomeric state representation of these polymers should permit estimation of the energies associated with conformations accessible to the skeletal bonds. Additionally, the conformation of these polymers in the crystalline state is known from detailed X-ray diffraction studies. It is therefore also of interest to compare this conformation with the conformation of lowest intramolecular energy, thereby gauging the effects, if any, of intermolecular interactions in the selection of a suitably regular chain conformation for adoption into a crystalline lattice. Theory Structure of the Repeat Units. On the basis of X-ray diffraction studies of crystalline cis-l,4-polybuta13,14 and extensive and cis-l,4-p01yisoprene~~~ data on low molecular weight analogs,'j the following = 1.53 A, structural parameters were adopted: lc=c = 1.34 A, Zc-H = 1.10 A, LCHZ-CH=CH = LCH2-C(CH3)==CH = 125" (e.g., 180" - 01, in Figure 1 and 180" - O2 in Figure 2), LCH-CH2CH2 = 112" (e.g., 180' - O3 and 180" - O4 in Figure I), LCH2-C--H = LCHZ-C-CH3 = 117.5" = 110" (e.g., O5 in Figure l), and LCH2-CH-H (e.g., Be in Figure 2). The X-ray crystallographic studies of cis- 1,4-polyisoprene by BunnI3 suggest a somewhat smaller value of -118" for LCHZ-C(CH~)=CH. This variation probably represents, at least in (11) G.Natta and P. Corradini, Angew. Chem., 68, 615 (1956). (12) G. Natta and P. Corradini, Nuooo Cimento, Suppl., 15, 1 1 1 (1960). (13) C.W.Bunn, Proc. Roy. SOC.(London), Al80, 40 (1942). (14) S.C. Nyburg, Acta Cryst., 7, 385 (1954). (15) H.J. M.Bowen and L. E. Sutton, "Tables of Interatomic Distances and Configurations in Molecules and Ions," The Chemical Society, London, 1958 ; "Interatomic Distances Supplement," The Chemical Society, London, 1965.

October 5, I966